Intrapartum epidural analgesia and breastfeeding: a prospective cohort study
© Torvaldsen et al. 2006
Received: 15 August 2006
Accepted: 11 December 2006
Published: 11 December 2006
Anecdotal reports suggest that the addition of fentanyl (an opioid) to epidural analgesia for women during childbirth results in difficulty establishing breastfeeding. The aim of this paper is to determine any association between epidural analgesia and 1) breastfeeding in the first week postpartum and 2) breastfeeding cessation during the first 24 weeks postpartum.
A prospective cohort study of 1280 women aged ≥ 16 years, who gave birth to a single live infant in the Australian Capital Territory in 1997 was conducted. Women completed questionnaires at weeks 1, 8, 16 and 24 postpartum. Breastfeeding information was collected in each of the four surveys and women were categorised as either fully breastfeeding, partially breastfeeding or not breastfeeding at all. Women who had stopped breastfeeding since the previous survey were asked when they stopped.
In the first week postpartum, 93% of women were either fully or partially breastfeeding their baby and 60% were continuing to breastfeed at 24 weeks. Intrapartum analgesia and type of birth were associated with partial breastfeeding and breastfeeding difficulties in the first postpartum week (p < 0.0001). Analgesia, maternal age and education were associated with breastfeeding cessation in the first 24 weeks (p < 0.0001), with women who had epidurals being more likely to stop breastfeeding than women who used non-pharmacological methods of pain relief (adjusted hazard ratio 2.02, 95% CI 1.53, 2.67).
Women in this cohort who had epidurals were less likely to fully breastfeed their infant in the few days after birth and more likely to stop breastfeeding in the first 24 weeks. Although this relationship may not be causal, it is important that women at higher risk of breastfeeding cessation are provided with adequate breastfeeding assistance and support.
Breastfeeding difficulties in the early postpartum period, especially if combined with inadequate support and advice, may lead new mothers to give up breastfeeding. Many factors influence the establishment of breastfeeding, including the type of labour and birth [1–4]. Local anecdotal reports were suggesting that the addition of fentanyl (an opioid) to epidural analgesia was resulting in difficulty establishing breastfeeding. Whilst the effects of epidurals on some obstetric consequences, including longer labour and an increased risk of an instrumental delivery , are well established, the impact, if any, of epidurals on breastfeeding is less clear.
A study conducted in Western Australia of 484 primiparous women with spontaneous vaginal births found epidural analgesia was associated with shorter breastfeeding duration (adjusted hazard ratio 1.44, 95% CI 1.04, 1.99) . Similarly, a study undertaken in Lapland of 64 primiparous women with spontaneous vaginal births found women who had epidural analgesia were more likely to be either partially breastfeeding or formula feeding at 12 weeks postpartum than women who had not had epidural analgesia (relative risk 2.27, 95% CI 1.27, 4.04) . Three of four small (56 ≤ n ≤ 189) North American studies examining the effects of epidurals on breastfeeding initiation and duration found no association [7–9], in spite of one study demonstrating an association between epidurals and reduced neonatal suckling scores . The fourth study (n = 108) found reduced rates of breastfeeding at six months among privately insured women with an epidural (30% vs 50%) .
This study aimed to determine whether there is any association between intrapartum epidural analgesia/anaesthesia and 1) breastfeeding in the first week postpartum and 2) breastfeeding cessation during the first 24 weeks postpartum.
This population-based, prospective cohort study has been described in detail previously . In brief, participants were women resident in the Australian Capital Territory (ACT) aged 16 years and over, who gave birth to a live baby between March and October 1997 in any of the ACT's two public hospitals (one included a birth centre), two private hospitals, or at home. Women were excluded if their baby was admitted to the neonatal intensive care unit or adopted, if critically ill themselves (i.e. too ill to be able to be asked to participate/give informed consent such as if they were admitted to the intensive care or psychiatric units or had an extended stay in delivery suite for medical reasons) or unable to give informed consent or complete the questionnaires for other reasons. For the analyses presented in this paper, women with multiple pregnancies were also excluded.
The study was approved by the ACT Department of Health & Community Care Research Ethics Committee, and the ethics committees of participating hospitals. Postnatal ward or domiciliary midwives gave information sheets to women in the first few days after giving birth. Women gave written informed consent when completing the first questionnaire as close to day four as possible (88% of respondents had completed the questionnaire by day 6). Women were then mailed questionnaires at 8, 16, and 24 weeks postpartum. The first questionnaire included sociodemographic characteristics of the mother and her partner, obstetric details and information on breastfeeding. Breastfeeding information was collected in each of the four surveys and women were categorised into one of three levels – fully breastfeeding (includes expressed breast milk), partially breastfeeding (breast milk plus some formula) or not breastfeeding at all. If women had stopped breastfeeding since the previous survey, they were asked when they stopped breastfeeding, to the nearest two weeks.
For each study participant, type of labour analgesia was classified into one of the following five mutually exclusive categories: 1) non-pharmacological only (natural methods such as breathing exercises, massage, moving about, hypnosis); 2) gas (nitrous oxide, no other pharmacological agent); 3) pethidine (+/- gas); 4) epidural (analgesia or anaesthesia, +/- pethidine +/- gas); or 5) general anaesthetic (+/- any of 1–4). The epidural solution in use at the time was bupivacaine 0.16% with fentanyl 3.3 μg/ml. This was given as patient controlled epidural analgesia with a 4–6 ml bolus and a 15 minute lockout (Dr Cliff Peady, Acting Director Anaesthetics, The Canberra Hospital, personal communication).
Associations between demographic and intrapartum factors and breastfeeding in the first week postpartum were investigated using χ2 tests. Odds ratios (OR) were estimated using logistic regression models to determine predictors of partial and no breastfeeding in the first week. For all models, variables that were individually significant at p < 0.05 were included initially; the variables with the highest p-value were then removed one at a time until only those with p < 0.05 remained. Some of the variables predictive of partial breastfeeding were on the causal pathway of others and so could not be included in a multivariate model. For example if there is no labour then the type of birth is caesarean section and the only analgesia options are epidural or general anaesthetic. Kaplan-Meier survival probability estimates were used to summarise time to cessation of breastfeeding during the first 24 weeks postpartum, among women who were partially or fully breastfeeding in the first week after birth. The hazard ratios (HR) of breastfeeding cessation in the first 24 weeks and their 95% confidence intervals (CI) of each exposure level were estimated by fitting a Cox proportional hazards regression model. All data analyses were undertaken using SAS version 8.2 (SAS Institute, Cary, NC).
Method of analgesia by demographic and obstetric factors
p-value from χ2test
n = 312
n = 190
n = 292
n = 416
n = 50
Up to year 11
Type of birth
CS with labour
CS no labour
Onset of labour
No previous births
≥1 previous births
Length of labour 1
Predictors of partial breastfeeding in the first postpartum week.
(compared with fully breastfeeding)
adjusted OR (95% CI)
1 (referent group)
1.46 (0.48, 4.45)
0.97 (0.34, 2.75)
3.61 (1.57, 8.29)
9.25 (3.22, 26.59)
1 (referent group)
2.59 (1.51, 4.45)
Type of birth 1
1 (referent group)
1.82 (0.91, 3.65)
CS* no labour
4.41 (2.25, 8.64)
CS with labour
6.22 (3.15, 12.30)
Onset of labour 1
1 (referent group)
1.86 (1.07, 3.22)
3.73 (1.92, 7.24)
Predictors of breastfeeding cessation in the first 24 weeks1
Crude HR (95% CI)
Adjusted HR2 (95% CI)
n/N (%) breastfeeding at 24 weeks
1 (referent group)
1 (referent group)
1.21 (0.84, 1.74)
1.20 (0.83, 1.74)
1.95 (1.45, 2.63)
1.67 (1.23, 2.25)
2.07 (1.57, 2.72)
2.02 (1.53, 2.67)
1 (referent group)
1 (referent group)
2.43 (1.95, 3.03)
2.21 (1.76, 2.78)
3.51 (2.65, 4.65)
2.67 (1.99, 3.58)
4.52 (2.60, 7.84)
3.19 (1.82, 5.61)
1 (referent group)
1 (referent group)
1.78 (1.32, 2.39)
1.67 (1.24, 2.26)
2.39 (1.84, 3.11)
2.01 (1.53, 2.64)
Up to year 11
2.97 (2.25, 3.91)
2.45 (1.84, 3.27)
To see if all the observed effect in the epidural group could have been due to the pethidine, we also undertook a Cox proportional hazards analysis using re-categorised analgesia groups. Instead of epidural +/- pethidine, we used epidural only and pethidine plus epidural, although it should be noted that all women in the epidural only group had a caesarean section. The HR for epidural only (adjusted for age and education) was 2.07 (95% CI 1.44, 2.96) and the HR for pethidine plus epidural was 1.73 (95% CI 1.30, 2.25).
In order to determine whether the increased rate of breastfeeding cessation among women who had an epidural was all due to those who were partially breastfeeding in the first week, we also undertook a Cox proportional hazards analysis of women who were fully breastfeeding in the first week who later reported either partially breastfeeding or not breastfeeding at all. Women who had an epidural during their labour and were fully breastfeeding in the first week were significantly more likely to stop fully breastfeeding than women who did not use any pharmacological agents during their labour (HR 1.59, 95% CI 1.25, 2.02); hence partial breastfeeding in the first week does not completely explain the increased breastfeeding cessation rate.
In this study, women who had epidural analgesia, a general anaesthetic and/or a caesarean section for the birth of their baby were more likely to be partially breastfeeding their baby in the first week postpartum. Partial breastfeeding in the first week may be indicative of difficulty establishing breastfeeding and is important because these women are at increased risk of stopping breastfeeding altogether. Two-thirds of the women who were partially breastfeeding in the first week were not breastfeeding at all by 24 weeks, compared with only one third of women who were fully breastfeeding in the first week. The type and amount of breastfeeding support given to these women is not known but may be an important factor in women continuing to breastfeed in spite of initial difficulties. International recommendations advise that women exclusively breastfeed their infants until six months of age and continue breastfeeding until at least 12 months of age .
Surprisingly, there was no overlap of predictors of partial and no breastfeeding in the first week. To the best of our knowledge, this has not been previously reported. The fact that no intrapartum factors were associated with not breastfeeding at all suggests that not breastfeeding has more to do with maternal preferences, which in turn are influenced by education. Women with a low level of education are at increased risk of both not initiating breastfeeding and, if they do breastfeed in the first few days, stopping in the first 24 weeks. Education has been previously reported to be associated with breastfeeding rates . Other previously reported risk factors for breastfeeding cessation, such as planning to return to work, were not significant in this cohort of women.
Intrapartum analgesia, type of birth and onset of labour are inextricably linked and it is hard to evaluate the relative contribution of each on breastfeeding. The effects of analgesia and type of birth are difficult to separate as each of these variables could be on the causal pathway of the other. When the analyses were restricted to women who had vaginal births, the strength of association between epidurals and partial breastfeeding and between epidurals and breastfeeding difficulties was weaker, and the association between epidurals and partial breastfeeding lost statistical significance. This suggests that at least some of the effect of epidurals could be due to its association with type of birth.
Although type of birth and parity were strong predictors of partial breastfeeding in the first week, neither was associated with breastfeeding cessation in the first 24 weeks postpartum. The only variable to be significantly associated with both partial breastfeeding in the first week and breastfeeding cessation during the first 24 weeks was intrapartum analgesia, with women in the epidural group being twice as likely to stop breastfeeding altogether in the first 24 weeks after birth as women who had non-pharmacological methods of analgesia during their labour. However, women stop breastfeeding for a variety of reasons, and this study did not collect information on all possible reasons.
Although none of the recorded potential confounders was significant, some or all of the association could be due to characteristics of the women who choose different sorts of analgesia. Women who choose non-pharmacological methods may be more likely to continue breastfeeding for at least the first 24 weeks and this could explain, either partly or wholly, the association between intrapartum analgesia and breastfeeding cessation. A previous study, in which women consented to be randomised to receive either an epidural or continuous midwifery care, also found women who had an epidural were more likely to stop breastfeeding in the first six months . Although this previous study had a crossover rate too high to permit a meaningful intention-to-treat analysis, women who felt strongly one way or the other about epidurals would be unlikely to consent to randomisation and so the likely confounding effect of personal choice would be reduced.
An alternative explanation for these results is that the association between breastfeeding and intrapartum analgesia is due to the pharmacological effect of the analgesic agents. There is a growing body of evidence that the fentanyl component of epidurals may be associated with sleepy infants and difficulty establishing breastfeeding. Lower neurologic and adaptive capacity scores have been reported among infants whose mothers had a bupivacaine and fentanyl epidural compared with mothers who had bupivacaine with sufentanil or bupivacaine alone . Another study found that mothers who had had intrapartum fentanyl were less likely to be breastfeeding at discharge from hospital and that there was a dose-response relationship between fentanyl and bottle feeding, independent of other factors . However the strongest evidence to date comes from a recent randomised controlled trial in which 177 women who had previously breastfed were randomised to receive an epidural containing either no fentanyl, intermediate dose fentanyl (up to 150 μg fentanyl) or high dose fentanyl (>150 μg fentanyl) . Women randomised to the high dose fentanyl group were more likely to have stopped breastfeeding at six weeks postpartum than women who were randomised to the other two groups . In our study we did not have information about the dosage or timing of the analgesic agents used for the individual women, so cannot draw any conclusions about this. The standard concentration of epidural fentanyl solution in the ACT at the time of this study was 3.3 μg/ml. This was lower than the usual concentrations used elsewhere in Australia and New Zealand at that time, which in many centres was 5 μg/ml, although this was variable. The usual dose now is 4 μg/ml, reflecting a slight increase in fentanyl concentration in the ACT but a reduction in concentration for some centres elsewhere (Dr Cliff Peady, Acting Director Anaesthetics, The Canberra Hospital, personal communication).
Whatever the underlying mechanism, women in this cohort who chose or needed epidural analgesia were more likely to partially breastfeed their infant and to experience difficulty breastfeeding in the few days after birth, and also to stop breastfeeding in the first 24 weeks postpartum. Even though this relationship may not be causal, it is important that women who are at higher risk of breastfeeding cessation are provided with adequate breastfeeding assistance and support, both in the initial postpartum period and the following few months.
We are grateful to Mary Lantry and Angela Smith (lactation consultants, Royal Prince Alfred Hospital for Women and Babies, Sydney, Australia) for raising the question of a possible association between epidurals and breastfeeding. Christine Roberts is supported by a National Health and Medical Research Council (NHMRC) of Australia Public Health Practitioner Fellowship and Siranda Torvaldsen is supported by a NHMRC Australian Research Training Fellowship. The cohort study was supported by a project grant from The Canberra Hospital Private Practice Fund. Additional funding was provided by The Canberra Hospital Auxiliary, the Nurses' Board of the Australian Capital Territory, and the Australian Capital Territory Department of Health & Community Care. The assistance of the midwives in recruiting women for this study is gratefully acknowledged. We also thank Marian Currie for her contribution to data collection. We are especially grateful to the women who so generously gave their time to complete this study.
- Sinusas K, Gagliardi A: Initial management of breastfeeding. American Family Physician 2001,64(6):981–988.PubMed
- Rowe-Murray HJ, Fisher JR: Baby friendly hospital practices: cesarean section is a persistent barrier to early initiation of breastfeeding. Birth 2002,29(2):124–131.View ArticlePubMed
- Jordan S, Emery S, Bradshaw C, Watkins A, Friswell W: The impact of intrapartum analgesia on infant feeding. BJOG: an International Journal of Obstetrics & Gynaecology 2005,112(7):927–934.View Article
- Henderson JJ, Dickinson JE, Evans SF, McDonald SJ, Paech MJ: Impact of intrapartum epidural analgesia on breast-feeding duration. Australian & New Zealand Journal of Obstetrics & Gynaecology 2003,43(5):372–377.View Article
- Anim-Somuah M, Smyth R, Howell C: Epidural versus non-epidural or no analgesia in labour. [http://www.cochrane.org/reviews/en/ab000331.html] Cochrane Database of Systematic Reviews 2005. Issue 4. Art. No.: CD000331. DOI: 10.1002/14651858.CD000331.pub2.
- Volmanen P, Valanne J, Alahuhta S: Breast-feeding problems after epidural analgesia for labour: a retrospective cohort study of pain, obstetrical procedures and breast-feeding practices. International Journal of Obstetric Anesthesia 2004,13(1):25–29.View ArticlePubMed
- Halpern SH, Levine T, Wilson DB, MacDonell J, Katsiris SE, Leighton BL: Effect of labor analgesia on breastfeeding success. Birth 1999,26(2):83–88.View ArticlePubMed
- Radzyminski S: The effect of ultra low dose epidural analgesia on newborn breastfeeding behaviors. J Obstet Gynecol Neonatal Nurs 2003,32(3):322–331.View ArticlePubMed
- Riordan J, Gross A, Angeron J, Krumwiede B, Melin J: The effect of labor pain relief medication on neonatal suckling and breastfeeding duration. Journal of Human Lactation 2000,16(1):7–12.View ArticlePubMed
- Kiehl EM, Anderson GC, Wilson ME, Fosson L: Social status, mother-infant time together, and breastfeeding duration. Journal of Human Lactation 1996,12(3):201–206.View ArticlePubMed
- Thompson JF, Roberts CL, Currie MJ, Ellwood DA: Early discharge and postnatal depression: a prospective cohort study. Med J Aust 2000,172(11):532–536.PubMed
- National Health and Medical Research Council: Dietary guidelines for children and adolescents in Australia - incorporating the infant feeding guidelines for health workers. [http://www.nhmrc.gov.au/publications/_files/n34.pdf] Commonwealth of Australia 2003.
- Scott JA, Aitkin I, Binns CW, Aroni RA: Factors associated with the duration of breastfeeding amongst women in Perth, Australia. Acta Paediatrica 1999,88(4):416–421.View ArticlePubMed
- Loftus JR, Hill H, Cohen SE: Placental transfer and neonatal effects of epidural sufentanil and fentanyl administered with bupivacaine during labor. Anesthesiology 1995,83(2):300–308.View ArticlePubMed
- Beilin Y, Bodian CA, Weiser J, Hossain S, Arnold I, Feierman DE, Martin G, Holzman I: Effect of labor epidural analgesia with and without fentanyl on infant breast-feeding: a prospective, randomized, double-blind study.[see comment]. Anesthesiology 2005,103(6):1211–1217.View ArticlePubMed
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.